Re: [volt-nuts] 3458A calibration
Hi all volt-nuts,
A quote from previous mails about 3458a calibration:
"What makes it all practical is its near perfect linearity and high
resolution".
Does anyone know how it is achieved? I read all that's available to
me and my main question is:
How to maintain this "perfect linearity". I read about multislope
ramp conversion, but that is a solution meant to shorten the
integration for such a big number of digits, not to improve linearity.
All sorts of effects come to my mind...: nonlinear active component
leakage, voltage coeficients in resistors (also resistive dividers),
nonlinear dielectric absorption in integration capacitors etc etc...
Does anyone has some 3458a AD schematics? Service manual can't be
downloaded from internet...
Predrag Dukic
In message 7.0.1.0.1.20090810173047.01d8cd90@tapko.de, Predrag Dukic writes:
How to maintain this "perfect linearity". I read about multislope
ramp conversion, but that is a solution meant to shorten the
Does anyone has some 3458a AD schematics? Service manual can't be
downloaded from internet...
The ADC in the HP3458A derives its linearity from a very small handful
of components, most importantly a quite small capacitor (~120pF),
by converting the measurement of voltage to a measurement of time.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Poul-Henning Kamp wrote:
In message 7.0.1.0.1.20090810173047.01d8cd90@tapko.de, Predrag Dukic writes:
How to maintain this "perfect linearity". I read about multislope
ramp conversion, but that is a solution meant to shorten the
Does anyone has some 3458a AD schematics? Service manual can't be
downloaded from internet...
The ADC in the HP3458A derives its linearity from a very small handful
of components, most importantly a quite small capacitor (~120pF),
by converting the measurement of voltage to a measurement of time.
The April 1989 HP journal article on the 3458A states that the
integrator uses a 330pF feedback capacitor.
The effect of dielectric absorption in the integrator feedback capacitor
is kept small by using a capacitor with low dielectric absorption and by
keeping the maximum charge stored in the capacitor a small fraction of
the full scale integrator input current integration time product.
The effect of dielectric absorption can be minimised by ensuring that
the average voltage across the integrator feedback capacitor remains low
over the integration cycle.
HP/Agilent have several patents covering this aspect of DVM integrator
runup cycles.
However the runup cycle of the 3458 unlike some other HP/Agilent DVMs
doesn't appear to use this technique.
Later HP DVMs do use such techniques.
Bruce
The effect of dielectric absorption can be minimised by
ensuring that the average voltage across the integrator feedback
capacitor remains low over the integration cycle.
HP/Agilent have several patents covering this aspect of DVM
integrator runup cycles.
Are you referring to what HP calls their "charge-balancing" A/D converter where during the run-up phase of the conversion they attempt to keep the output of the integrator balanced around 0 volts?
The Agilent 34401A does something like this and then uses a second, fast A/D converter to replace the run-down phase.
However the runup cycle of the 3458 unlike some other
HP/Agilent DVMs doesn't appear to use this technique.
Later HP DVMs do use such techniques.
It would seem that with an A/D, like the 3458A's, that relies on multi-slope run-down, it would be advantageous to keep the integrator output as high as possible at the end of the run-up phase to maximize the resolution achieved during run-down.
Randy.
Randy Scott wrote:
The effect of dielectric absorption can be minimised by
ensuring that the average voltage across the integrator feedback
capacitor remains low over the integration cycle.
HP/Agilent have several patents covering this aspect of DVM
integrator runup cycles.
Are you referring to what HP calls their "charge-balancing" A/D converter where during the run-up phase of the conversion they attempt to keep the output of the integrator balanced around 0 volts?
The Agilent 34401A does something like this and then uses a second, fast A/D converter to replace the run-down phase.
The 34401 runup algorithm is a little more complex than that, it
actually sums the output of the integrator with the input to the
integrator and uses that as the comparator input.
The term balanced is ambiguous, I mean that the algorithm attempts to
keep the integral of the integrator output constant.
Even lower dielectric absorption effects are achieved if a term
proportional to the integral of the integrator output is included in
effect a second order delta sigma runup cycle.
However the runup cycle of the 3458 unlike some other
HP/Agilent DVMs doesn't appear to use this technique.
Later HP DVMs do use such techniques.
It would seem that with an A/D, like the 3458A's, that relies on multi-slope run-down, it would be advantageous to keep the integrator output as high as possible at the end of the run-up phase to maximize the resolution achieved during run-down.
Randy.
Bruce
Keeping the voltage across the integrating capacitor (0 is a special case
of constant) is a good idea because it solves the problem of dC/dV
(voltage coefficient of capacitance).
-John
============
The effect of dielectric absorption can be minimised by
ensuring that the average voltage across the integrator feedback
capacitor remains low over the integration cycle.
HP/Agilent have several patents covering this aspect of DVM
integrator runup cycles.
Are you referring to what HP calls their "charge-balancing" A/D converter
where during the run-up phase of the conversion they attempt to keep the
output of the integrator balanced around 0 volts?
[snip]
For the 3458A integrator feedback capacitor nonlinearity isnt an issue
(as long as there is no hysteresis) as it cancels out during the rundown
phase.
For DVMs like the 34401A which use an ADC to sample the integrator
output integrator feedback capacitor nonlinearity is important.
Bruce
J. Forster wrote:
Keeping the voltage across the integrating capacitor (0 is a special case
of constant) is a good idea because it solves the problem of dC/dV
(voltage coefficient of capacitance).
-John
============
The effect of dielectric absorption can be minimised by
ensuring that the average voltage across the integrator feedback
capacitor remains low over the integration cycle.
HP/Agilent have several patents covering this aspect of DVM
integrator runup cycles.
Are you referring to what HP calls their "charge-balancing" A/D converter
where during the run-up phase of the conversion they attempt to keep the
output of the integrator balanced around 0 volts?
[snip]
Re: Fluke p/n 739961 6V 4.5AH lead-acid batteries for Fluke 732A DC
Reference Standard
Does anyone know of an affordable substitute for these batteries? Fluke's
web site doesn't even recognize their own part number for these batteries,
but Newark lists them at $67.49 each. The unit uses qty 4, so at Newark's
price a set of four would cost $269.96!
According to Newark their physical size is 70x45x107mm.
The old set that were in my 732A are marked, "PT56 6V 5AH Made in China". So
someone installed these as substitutes. I searched the internet for "PT56"
but didn't see anything linking that number to a standard or generic
battery.
Thanks,
Greg
Try a place that sells to professional alarm installers.
-John
=============
Re: Fluke p/n 739961 6V 4.5AH lead-acid batteries for Fluke 732A DC
Reference Standard
Does anyone know of an affordable substitute for these batteries? Fluke's
web site doesn't even recognize their own part number for these batteries,
but Newark lists them at $67.49 each. The unit uses qty 4, so at Newark's
price a set of four would cost $269.96!
----- Original Message -----
From: "J. Forster"
Try a place that sells to professional alarm installers.
-John
=============
OK, good suggestion. I'll try that.
Thanks,
Greg